Journal of Hazardous Materials 213214 (2012) 242248

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials

journal homepage: www.elsevier.com/locate/jhazmat

Optimization and kinetic modeling of cadmium desorption from citrus peels: A process for biosorbent regeneration
Eloh Njikam, Silke Schiewer
Department of Civil and Environmental Engineering, University of Alaska Fairbanks, P.O. Box 755900, Fairbanks, AK 99775, United States

a r t i c l e

i n f o

a b s t r a c t
Citrus peel biosorbents are efcient in removing heavy metals from wastewater. Heavy metal recovery and sorbent regeneration are important for the nancial competitiveness of biosorption with other processes. The desorbing agents HNO3 , NaNO3 , Ca(NO3 )2 , EDTA, S, S-EDDS, and Na-Citrate were studied at different concentrations to optimize cadmium elution from orange or grapefruit peels. In most cases, desorption was fast, being over 90% complete within 50 min. However sodium nitrate and 0.001 M nitric acid were less efcient. Several new models for desorption kinetics were developed. While zero-, rst- and second-order kinetics are commonly applied for modeling adsorption kinetics, the present study adapts these models to describe desorption kinetics. The proposed models relate to the number of metal-lled binding sites as the rate-determining reactant concentration. A model based on rst order kinetics with respect to the remaining metal bound performed best. Cd bound in subsequent adsorption after desorption was similar to the original amount bound for desorption by nitric acid, but considerably lower for calcium nitrate as the desorbent. While complexing agents were effective desorbents, their cost is higher than that of common mineral acids. Thus 0.010.1 M acids are the most promising desorbing agents for efcient sorbent regeneration. 2012 Elsevier B.V. All rights reserved.

Article history: Received 26 October 2011 Received in revised form 25 January 2012 Accepted 26 January 2012 Available online 2 February 2012 Keywords: Biosorption Citrus peels Desorption Kinetics Modeling

1. Introduction Heavy metal ions such as cobalt, copper, nickel, chromium, cadmium, and zinc are present in the waste streams from mining operations, tanneries, textile mills, electronics, electroplating and petrochemical industries [1]. Heavy metals have a harmful effect on human physiology and other biological systems when they exceed the tolerance levels. They are not biodegradable and tend to accumulate in living organisms, causing various diseases and disorders. Conventional methods used to remove heavy metals from waste water (precipitation, ion exchange, redox processes, reverse osmosis, etc.) often involve high capital and operating cost, especially when the metals are dissolved in large volumes of solution at relatively low concentrations (around 1100 mg/L), and when high efuent quality is desired [2]. Studies are thus turning towards costeffective methods. Biosorption, using biosorbents from industrial or natural sources, provides an efcient and competitive solution to treat wastewater, particularly for high volumes of dilute solutions [3].

Corresponding author. Tel.: +1 907 474 2620; fax: +1 907 474 6087. E-mail addresses: ennjikam@alaska.edu (E. Njikam), sschiewer@alaska.edu (S. Schiewer). 0304-3894/$ see front matter 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2012.01.084

Biosorption can be dened as the sequestering of metals by materials of biological origin. Biosorption is metabolismindependent such that non-living biomass can act as a biosorbent [4]. Metal ion binding may involve complexation, coordination, chelation, ion exchange, adsorption and inorganic microprecipitation depending on the type of biosorbent [5]. Recent studies have determined biosorption using citrus peels as being efcient in the removal of heavy metals from wastewater with sorption capacities of 0.351.15 mmol/g (39128 mg/g) of Cd per biosorbent dry weight [6,7]. The economic feasibility and subsequent advantage of this process over others will depend on the amount of metal bound by the biosorbent and the ability to meet legal wastewater discharge standards and biosorbent regeneration for use in multiple adsorptiondesorption cycles [8,9]. The desorption mechanism is similar to that of biosorption and may involve ion exchange or complexation, where metals are eluted from the biosorbent by an appropriate solution to produce a small, concentrated volume of metal-containing solution [10], from which metals could potentially be recovered. Studies have shown that the protons of mineral acids such as HCl, H2 SO4 and HNO3 can displace metals from the sorbents binding sites [8,11] and that the strong chelating agent EDTA is also an efcient desorbing agent [12,13], causing metal desorption by reducing the concentration of free metal ions in solution. Taking the above points into consideration, and considering the fact that citrus peels are efcient in

E. Njikam, S. Schiewer / Journal of Hazardous Materials 213214 (2012) 242248

243

the removal of cadmium from waste water, one objective of this study was to examine the efciencies of selected desorbing agents (HNO3 , NaNO3 , Ca(NO3 )2 , NaCitrate, Na2 EDTA, and S,S-EDDS) at varying concentrations in order to determine the most economic highly effective desorption agent. Since there are currently no published studies on modeling desorption kinetics, with the notable exception of Singh et al. who described Cu2+ and Pb2+ desorption by 0.1 M HCl using 1st and 2nd order Lagergren models [14], a second objective was to develop models that are able to describe desorption kinetics for a variety of desorbents. The third objective was to verify efcient sorbent regeneration for the chosen desorbing agent by conducting isotherm experiments in order to quantify Cd bound by the previously desorbed peels. 2. Materials and methods 2.1. Materials Dried peels of two different citrus varieties, processed Valencia orange peels (Alarma Consulting Corporation, FL) and waste grapefruit (Citrus paradise) peels from juice production (Peace River Citrus Products Company, FL), were ground and sieved to a 11.1 mm size diameter. Protonation of the peels was carried out in order to remove light metal cations such as Na or Ca, and to saturate binding sites such as carboxyl groups with protons, according to a modied procedure by Schiewer and Balaria [15]. The sorbent material was mixed with 0.1 M HNO3 (20 g of peels/L) for 240 min on a magnetic stirrer, rinsed with nanopure water (approximately 18 resistance), and dried for 720 min at 40 C. The weight of the peels before and after protonation was recorded and weight loss was found to be negligible. 2.2. Adsorption isotherm Adsorption isotherm studies were carried out by varying the initial metal concentrations from 0.089 to 8.89 mM (10 mg/L to 1000 mg/L). A 0.05 L solution of the desired cadmium (Cd2+ ) concentration was prepared, using 89 mM (10,000 mg/L) AA (atomic absorption) standard solution of cadmium nitrate to prevent complexation. 0.05 g of the dried protonated citrus peels were put in contact with the prepared solutions in Erlenmeyer asks placed on a rotary shaker for an equilibration time of 180 min. The pH was controlled and adjusted to 5.0 during the contact period, using 0.1 M NaOH. The samples were ltered on Whatman 40 ashless lter paper and the ltrate acidied with 103 L of concentrated (16 M) nitric acid, and stored (at 3 C) for analysis on the Atomic Absorption Spectrometer (Perkin Elmer AAnalyst 300). 2.3. Desorption studies All desorption experiments were carried out using the batch process. Initial loading of cadmium onto the citrus peels was done by exposing 20 g of the dried protonated citrus peel to 1 L of 4.46 mM Cd(NO3 )2 solution on a magnetic stirrer for 180 min (at pH 5). The amount of Cd loaded onto the citrus peels was 0.23 mmol/g for the grapefruit peels, and 0.305 mmol/g for the orange peels. Cd2+ loaded orange peels were used for the desorption experiments with the desorbents NaCitrate, Na2 EDTA, and S,SEDDS (desorption by Cd2+ complexation in solution), and Cd2+ loaded grapefruit peels were used for the desorption experiments with HNO3 , NaNO3 , Ca(NO3 )2 as desorbents (desorption by ion exchange). Nitrate was chosen as the anion for that set of experiments since nitrate does not form complexes with Cd2+ under the studied conditions. Kinetic studies were conducted in order to determine the desorption rate of the heavy metal from the citrus peels. The desorption

studies were done by contacting 0.05 g of the loaded biomass with 0.05 L of the different desorbing agents at the chosen concentrations (0.001, 0.01, 0.1, and/or 1 M) on a rotary shaker. At different contact times (51440 min), the samples were immediately ltered on Whatman 40 ashless lter paper and stored for analysis. The data collected were modeled using different kinetic models. 2.4. Theoretical background 2.4.1. Adsorption isotherms The metal bound per gram of biosorbent material was determined using the mass balance. If C0 and Cf are the initial and nal metal concentrations (mol/L), respectively, V is the suspension volume (L) and m is the mass of biosorbent material (g), then the metal bound q (mmol/g) can be calculated as: q= V (C0 Cf ) m (1)

The results were plotted as sorption isotherms (Langmuir, Freundlich, and BET) of the sorbate bound (q) versus the equilibrium concentration of the residual sorbate remaining in the solution (Cf ). The Langmuir adsorption model is valid for single-layer adsorption and assumes that all the binding sites are free sites. The Langmuir isotherm equation is: q = qmax KCf 1 + KCf (2)

where qmax is the maximum sorbate bound under the given conditions [mmol/g] and K [g/mmol] is a coefcient related to the afnity between the sorbent and sorbate. The Freundlich isotherm model is the empirical equation: q = kCf 1/n (3)

where k and n are the Freundlich constants. The BET model represents sorption isotherms reecting apparent multi-layer adsorption. It assumes that a number of layers of adsorbate molecules form at the surface. For low sorbate concentrations, where the rst layer is not yet saturated, the model approaches the Langmuir equation. The BET isotherm is given by the equation: q= Kqmax Cf (Cs Cf )[1 + (K 1)(Cf /Cs )] (4)

where Cs is the saturation concentration of the solute; K is a constant related to the energy of interaction with the surface and qmax is the moles of solute adsorbed per unit weight of adsorbent forming a complete monolayer on the surface. 2.4.2. Kinetic models 1st and 2nd order models are commonly used for adsorption, assuming that the adsorption rate is related to the difference between amount of metal bound at any given time versus the metal bound at equilibrium. To provide a model for desorption kinetics, which have so far not been modeled in the literature, the present work proposes to modify the commonly used kinetic models for adsorption and adapt them to desorption, with the remaining amount of metal bound being the rate-determining concentration. The desorption data obtained for the different desorbing agents were modeled according to a modication of the Lagergren zero-order, pseudo rst and second order adsorption kinetic model. The modied models proposed in this paper consider the metal-biosorbent desorption reaction as the rate-limiting step. They regard the metal-saturated sites at the biosorbent surface as the reactant and are based on different assumptions of how the desorption rate depends on the quantity of these cadmium-lled binding sites.

244

E. Njikam, S. Schiewer / Journal of Hazardous Materials 213214 (2012) 242248

dq = k0,des dt

(5)

Cd bound q (mmol/g)

The modied 0-order model assumes that the rate of desorption is constant, and independent of the quantity of the Cd2+ lled sites. It can be expressed as:

3.0 2.5 2.0 1.5 1.0 0.5 0.0 0 1 2

Data Langmuir Freundlich BET

where q (mmol/g) is the remaining amount of metal bound to the biosorbent at any time t, dq/dt is the desorption rate, and k0,des the 0-order desorption rate constant (mmol g1 min1 ). The pseudo 1st-order model assumes that the rate of desorption is proportional to the number of Cd2+ lled sites and can be expressed as: dq = k1,des q dt (6)

where k1,des is the 1st-order desorption rate constant (min1 ). The modied pseudo-2nd-order model assumes that the rate of desorption is proportional to the square of the number of Cd2+ lled sites and can be expressed as: dq = k2,des q2 dt (7)

Cadmium Concentration Cf (mM)

Fig. 1. Adsorption isotherm for cadmium binding by grapefruit peels (1 mm diameter) at pH 5. Data and predictions of Langmuir, Freundlich and BET isotherms. Langmuir and Freundlich model parameters were determined from data points up to 3.5 mM.

where k2,des is the 2nd-order desorption rate constant (g mmol1 min1 ). Stepwise numerical integration over time was performed for all three rate expressions according to the following equation: qn+1 = qn dq dt (tn+1 tn ) (8)

where qn is the amount of Cd remaining in peels at time tn and q0 is the initial amount of Cd in the peels (at time t0 = 0), as determined from a prior adsorption step. From the equation we see that the amount of Cd2+ (mmol/g) found in the peels at time tn+1 will be the amount at time tn minus the rate of desorption multiplied by the incremental time step. 2.4.3. Error analysis Error analysis was done in order to evaluate the errors of the model predictions of both the isotherm and desorption kinetic models. The root mean square error (RMSE) was calculated with the equation:
p

(q qm )2 RMSE =
1

(9)

where q is the experimental metal binding data, qm is the model predictions of metal bound, and p is the number of data points. The parameters for kinetic models and isotherms were optimized using the solver function in Excel by minimizing the RMSE values. 3. Results and discussion 3.1. Adsorption isotherm The optimization of Cd2+ desorption from loaded citrus peels began with characterizing the maximum amount of Cd2+ bound by grapefruit peels by creating a sorption isotherm for cadmium. These data were later used to compare to cadmium biosorption by previously desorbed peels, to determine whether regenerated peels were similarly efcient as fresh peels. The metal bound by the grapefruit peels as a function of the equilibrium cadmium concentration is shown in Fig. 1. The data showed a steep initial increase in metal bound with increasing concentration. The slope of the curve gradually decreased and then showed a sharp increase beyond the metal

binding of 1 mmol/g (at the nal metal concentration of 4.20 mM). The data showed a resemblance to the BET model. The Langmuir, Freundlich and BET models were tted to the data by nonlinear parameter optimization using the Excel Solver function such that the RMSE was minimized. Since the Langmuir and Freundlich models did not t well for the data at high concentrations, Table 1 shows not only model parameters obtained by tting the model over the entire data range, but also model parameters obtained by tting only to the data points up to 3.5 mM. For those low cadmium concentrations (Cf = 0.0023.5 mM), the Freundlich isotherm had the best t (lowest RMSE value). However when considering all data points, the BET isotherm had by far the lowest RMSE value. The results obtained at low cadmium concentrations are in accordance with a similar study by Schiewer and Patil [6] on the use of pectin-rich fruit wastes as biosorbents, where the Freundlich and Langmuir isotherms both provided a good t for adsorption of cadmium by different citrus peels at pH 5 and pH 3. No conclusion could be drawn about the underlying mechanism responsible for the observed increase in Cd binding at higher concentrations, which resulted in a better t of the BET model. Other biosorption studies concluded that multilayer sorption occurred when data followed a BET type isotherm [16,17]. In some cases, the BET isotherm and corresponding multilayer sorption was invoked even though the data did not follow a typical BET isotherm behavior (Langmuir type plateau followed by isotherm becoming steeper as asymptotic value is approached) [1820]. The authors of the present study are wary of over-interpreting results. Multilayer adsorption, which is commonly associated with the BET isotherm, would be highly unlikely due to a lack of binding sites for Cd 2+ ions once the rst sorbatesorbent layer is formed. No cadmium precipitation was observed even at higher cadmium concentrations. Visual Minteq (Ver. 2.61) was used to calculate the equilibrium speciation and evaluate whether precipitation of Cd complexes could have occurred. The Minteq modeling showed that even at the highest cadmium concentration of 7 mM, over 97% of Cd was present as Cd2+ , with 2.4% of Cd present as CdNO3 + and virtually no hydroxides. No formation of precipitate would have occurred according to Minteq. A possible explanation for the sudden increase in Cd2+ bound observed after Cf = 3.5 mM is a change in the stoichiometry of the adsorption model from a 1:2 stoichiometry (where one divalent metal ion binds to two monoproctic sites) to a 1:1 stoichiometry (one divalent metal ion binds to one monoproctic site).

E. Njikam, S. Schiewer / Journal of Hazardous Materials 213214 (2012) 242248 Table 1 Equilibrium parameters of LangmuirFreundlich and BET isotherm models for cadmium bound by grapefruit peels at pH 5. Isotherm Langmuir Freundlich BET
a b

245

Model parameters K (L/mmol) k K () 1.7 0.0027b 0.66a 0.44b 51a 70b

a

RMSE (mmol/g) qmax (mmol/g) n qmax (mmol/g) 1.13 126b

a

0.092a 0.231b 0.053a 0.223b Cs (mM) 10.6a 8.68b 0.081a 0.073b

2.83a 1.22b 0.72a 0.63b

Using only data with Cf < 3.5 mM to determine parameters. Using all data to determine parameters.

3.2. Desorption efciency and kinetics The desorption efciency and kinetics studies for the selection of optimal elution conditions were done using varying concentrations of several desorbents based on their efciency in Cd2+ removal from the citrus peels. For each desorbent and at each concentration, the desorption efciency DE (%) was calculated as: DE = q0 q q0 100 (10)

sludge, and grape stalks using these complexing agents [12,13,21]. As seen in Fig. 3, all three complexing agents were highly efcient, achieving 100% desorption. Efcient desorption by EDTA was also noted for other agricultural waste materials [22]. With the exception of 0.001 M EDTA, they all showed fast kinetics, achieving 90% desorption within the rst 50 min. Even after long exposure to the complexing agents, desorption remained at 100%, unlike observations for acidic desorption. Apparently, complexing agents did not damage the citrus peels. 3.3. Desorption models The desorption kinetics of the different desorbents were modeled according to the zero, 1st and 2nd order models, whose

where q0 is the amount of metal (Cd2+ ) bound into the citrus peels before desorption and q is the amount of remaining metal bound after desorption (mmol/g). The desorption efciency and kinetics for the different concentrations of the selected desorbents HNO3 , NaNO3 , Ca(NO3 )2 (desorption by ion exchange with competing ions pushing Cd2+ off binding sites) as well as Na-Citrate, Na2 EDTA, and S,S-EDDS (desorption by complexing agents pulling Cd2+ into solution) can be seen in Figs. 2 and 3, respectively. A logarithmic time scale was chosen to achieve a better resolution for the rst 30 min, during which much of the sorption took place. As seen in Fig. 2, the concentrations used were 1 M, 0.1 M, 0.01 M and 0.001 M for HNO3 , 1 M and 0.1 M for Ca(NO3 )2 , and 1 M for NaNO3 . For each desorbing agent, concentrations were lowered until less than 70% desorption was achieved in 30 min. Therefore, lower concentrations were investigated for the more efcient HNO3 compared to the weaker desorbing agent NaNO3 . With the exception of NaNO3 and 0.001 M HNO3 , all desorbent solutions were able to achieve 90100% removal of cadmium from the citrus peels within 120 min or less. HNO3 at 1 M, 0.1 M and 0.01 M showed a fast desorption rate, achieving 90% removal within the rst 60 min. Surprisingly, 1 M HNO3 showed a slightly lower efciency than 0.1 M and 0.01 M. This was most likely due to the fact that the high acid concentration damaged the overall structure of the citrus peels, also indicated by a discoloration of the solution. This was also noticed for 0.1 M HNO3 , which showed a reduction in efciency after 240 min. The 0.01 M HNO3 showed optimum performance, achieving 90% removal within the rst 50 min without damaging the peels. Overall, Ca(NO3 )2 displayed a lower desorption rate than HNO3 . Even for 1 M Ca(NO3 )2 desorption was lower than for 0.01 M HNO3 , i.e. concentrations for Ca(NO3 )2 had to be about 100 times higher than HNO3 concentrations to have the same effect. For the rst 30 min, desorption was faster for 1 M Ca(NO3 )2 than for 0.1 M Ca(NO3 )2 , after that, both concentrations showed similar results. NaNO3 was the weakest of the desorbing agents compared in Fig. 2, 1 M NaNO3 performed similar as 0.001 M HNO3 , i.e. the concentration had to be a thousand times higher for NaNO3 than for HNO3 . Fig. 3 shows the kinetics and efciency of the different complexing agents whose concentrations were 0.1 M and 0.001 M for EDTA, 0.001 M for S,S-EDDS and 0.1 M for Na citrate, based on previous studies on the desorption of cadmium or lead from algae, activated

a 110
100 90 80 70 60 50 40 30 20 10 0 1 10 100
Time (min) Desorption Efficiency DE (%)

1 M HNO3 0.1 M HNO3 0.01 M HNO3 0.001 M HNO3

1000

10000

b
Desorption Efficiency DE (%)

110 100 90 80 70 60 50 40 30 20 10 0 1 10 100

0.1 M HNO3 1 M CaNO3 0.1 M CaNO3 1 M NaNo3

1000

10000

Time (min)
Fig. 2. Desorption kinetics using different desorbing agents for grapefruit peels with an initial amount of Cd2+ bound of 0.23 mmol/g. (a) Comparison of different concentrations of HNO3 . (b) Comparison of HNO3 , Ca(NO3 )2 , and NaNO3 .

246

E. Njikam, S. Schiewer / Journal of Hazardous Materials 213214 (2012) 242248

Table 2 Parameters for 0, 1st and 2nd order kinetic models of selected desorbents for Cd elution from citrus peels. Desorbent Rate constant k Zero (mmol g HNO3 1 Ma HNO3 0.1 Ma HNO3 0.01 Ma HNO3 0.001 Ma Ca(NO3 )2 1 Ma Ca(NO3 )2 0.1 Ma NaNO3 1 Ma EDTA 0.1 Mb EDTA 0.001 Mb SS,EDDS 0.001 Mb NaCitrate 0.1 Mb
a b 1

RMSE min
1

1st (min ) 0.113 0.103 0.054 0.005 0.044 0.026 0.003 0.102 0.063 0.041 0.056

2nd (g mmol 0.702 0.672 0.434 0.029 0.338 0.196 0.022 0.474 0.088 0.228 0.298

min

Zero (mmol/g) 0.131 0.132 0.108 0.031 0.095 0.074 0.028 0.189 0.080 0.136 0.156

1st (mmol/g) 0.014 0.005 0.009 0.024 0.012 0.016 0.024 0.020 0.060 0.019 0.016

2nd (mmol/g) 0.016 0.019 0.014 0.016 0.011 0.024 0.030 0.033 0.011 0.021 0.024

0.0014 0.0018 0.0016 0.0007 0.0015 0.0015 0.0005 0.0025 0.0015 0.0020 0.0020

Grapefruit peels with q0 = 0.23 mmol/g. Orange peels with q0 = 0.305 mmol/g.

parameters are shown in Table 2. The parameter k was determined by non-linear parameter optimization, with the root mean square error (RMSE) being minimized using the solver function in Excel. As described above, the experimentally determined initial amount of Cd2+ in the peels before desorption was used in the desorption model, i.e. q0 = 0.23 mmol/g for the grapefruit peels with HNO3, NaNO3, Ca(NO3 )2 , and q0 = 0.305 mmol/g for the orange peels with Na-Citrate, Na2 EDTA, and S,S-EDDS. The data used for tting the kinetic models did not include the data points where severe disintegration occurred (the last data point at 1440 min for the desorption using 1 M and 0.1 M HNO3 ), as disintegration of the peels is a secondary process that masks the actual desorption kinetics. As Fig. 4 shows, rate constants increased with increasing desorbent concentration, i.e. desorption was faster at higher concentrations (except for 1 mM for zero order kinetics, which however was not an appropriate model assumption for the data set). For 0.1 mM, the 1st and 2nd order rate constants were already maximal; it would therefore be nancially wasteful to use higher concentrations. It may overall even be most benecial to use the more economical 0.01 mM HNO3 in spite of slightly lower rate constants. 0.1 mM EDTA was similarly efcient as 0.1 mM HNO3 , with 1st order rate constants k1,des 0.10 min1 . Among the complexing agents, SS,EDDS was stronger (higher rate constant) than EDTA, which in turn was stronger than Na-Citrate at the came concentration. The data in Table 2 show that the desorption kinetics data t well to both the rst and second order kinetic models, with RMSE of 0.010.05 mmol/g for the best tting model, which is small compared to the initial cadmium bound of 0.23 mmol/g for grapefruit

peels and 0.305 mmol/g for orange peels. With the exception of 0.001 M HNO3 , 0.001 M EDTA, and 1 M Ca(NO3 )2 , for which the 2nd order model t best, all desorbents showed a slightly better t for the rst order model. The close t of the two models may be either due to the fact that the selected concentrations of the reactant species (desorbents) are not in the rate-governing range, and initial excess of any of the reactants will give a mixed response or because the order of desorption is a fractional value between the rst and second order [14]. A comparison of the different order models to the desorption data can be seen in Fig. 5 which shows the performance of all the three rate models for 0.01 M and 0.001 M HNO3 . Those were chosen as representative examples where the 1st order model t well (as in most cases) and where the 1st order model t was less satisfactory, which was the case for 0.001 M HNO3 because of incomplete desorption. Current research focuses on alternative models for such cases. 3.4. Adsorption by regenerated peels The success of citrus peel regeneration after desorption with 0.1 M HNO3 and 0.1 M Ca(NO3 )2 was also studied. Those desorbents and their concentrations were chosen as being favorable based on their cost, effect on the environment, desorption rate, efciency, and the effect on the peels. The cadmium biosorption isotherms

0.8 desorption rate const. k1, k2 0.7 0.6 0.5 0.4 0.3 0.2 0.1

110 100 90 80 70 60 50 40 30 20 10 0 1 10 100

Desorption Efficiency DE (%)

0.003

0.002

0.1 M EDTA 0.1 M NaCitrate 0.001 M S,S-EDDS 0.001 M EDTA

0.001

1000

10000

Time (min)
Fig. 3. Comparison of desorption kinetics using S,S-EDDS, Na-Citrate and EDTA for orange peels with an initial amount of Cd2+ bound of 0.305 mmol/g.

0.0 0 0.01 0.1 1 0.001 HNO3 concentration (mM)

Fig. 4. Effect of HNO3 concentration on zero, 1st and 2nd order rate constant for cadmium desorption from grapefruit peels with an initial amount of Cd bound of 0.23 mmol/g.

desorption rate constant k 0

2nd order 1st order 0 order

0.004

E. Njikam, S. Schiewer / Journal of Hazardous Materials 213214 (2012) 242248

247

a 0.25
Cd bound, q (mmol/g) 0.20 Data 0 order 1st order 2nd order

0.15 0.10

0.05 0.00 0 50 100 150 Time (min) 200 250

strong competitors for the same binding sites. Calcium has been shown to repress cadmium and zinc binding, respectively, by seaweeds in biosorption experiments where both metals were present at varying concentrations [23,24]. Preliminary results of ongoing experiments of the authors similarly show cadmium binding by citrus peels decreasing with increasing calcium concentration. While in the current study HNO3 was used to be able to distinguish clearly between on the one hand competitive desorption by H+ or other cations that push Cd2+ off the binding sites and on the other hand desorption by complexing agents which pull Cd2+ into the solution, in industrial applications other mineral acids such as HCl, which has a low cost and is of lower environmental concern than some other acids, may be used.

4. Conclusions Several new models based on zero-, rst- and second-order kinetics were adapted here to describe desorption kinetics. The models relate to the number of cadmium-lled binding sites as the rate-determining reactant. The rst order kinetic desorption model t data for Cd desorption from citrus peels well, except for those cases where desorption remained incomplete, namely 0.001 M HNO3 and 1 M NaNO3 . The most promising desorbents were identied, considering criteria such as desorption effectiveness, desorption rate, cost, environmental effect and reuse-potential. Calcium salt was effective for desorption but slower than mineral acid and resulting in lower metal binding of Ca-regenerated peels compared to original or acid-regenerated peels. Mineral acids and complexing agents were fast and effective, achieving 90% desorption in less than 60 min. Since complexing agents are more expensive than common mineral acids, and 0.001 M acid was not an effective desorbent, 0.1 or 0.01 M acid were identied as the best desorbents. While HNO3 was used in this study, other mineral acids such as the relatively harmless and cheap HCl are promising as industrial desorbents.

b
Cd bound, q (mmol/g)

0.25 0.20 0.15 0.10 0.05 0.00 0 50 100 150 Time (min) 200 250 Data 0 order 1st order 2nd order

Fig. 5. Kinetics of cadmium desorption by HNO3 for grapefruit peels with an initial amount of Cd bound of 0.23 mmol/g. Data and predictions of zero, rst and second order model. (a) 0.01 M HNO3 and (b) 0.001 M HNO3 .

by the previously loaded citrus peels after desorption using 0.1 M HNO3 and 0.1 M Ca(NO3 )2 can be seen in Fig. 6. The cadmium bound by the peels in an adsorption step after desorption using 0.1 M HNO3 was very similar to that of the original peels. On the other hand the cadmium bound after desorption using 0.1 M Ca(NO3 )2 was much lower. This was most likely due to the fact that the two divalent ions cadmium and calcium are

Acknowledgments This research was supported by National Research Initiative of the USDA Cooperative State Research, Education, and Extension Service, grant number 2005-35504-16092 and by the USGS NIWR program. The authors are grateful to Robert Jones of Alarma Consulting Corp., Florida, and Peace River Citrus Products for donating citrus peels. None of the funding sources played a role in the design or execution of this study or the interpretation of its results.

1.6 1.4

Cd bound, q (mmol/g)

Original Desorbed by 0.1 M HNO3 Desorbed by 0.1 M Ca(NO3)2

References
[1] J.W. Patterson, R. Passino, Metals Speciation, Separation and Recovery, Lewis Publishers, Chelsea, MI, 1987. [2] E. Valdman, S.G.F. Leite, Biosorption of Cd, Zn and Cu by Sargassum sp. waste biomass, Bioprocess. Eng. 22 (2) (2000) 171173. [3] B.S. Wilhelmi, J.R. Duncan, Metal recovery from Saccharomyces-cerevisiae biosorption columns, Biotechnol. Lett. 17 (9) (1995) 10071012. [4] V.M. Dronnet, C.M.G.C. Renard, M.A.V. Axelos, J.-F. Thibault, Binding of divalent metal cations by sugar-beet pulp, Carbohydr. Polym. 34 (1-2) (1997) 7382. [5] B. Volesky, Biosorption of Heavy Metals, CRC Press, Boca Raton, FL, 1990. [6] S. Schiewer, S.B. Patil, Pectin-rich fruit wastes as biosorbents for heavy metal removal: equilibrium and kinetics, Bioresour. Technol. 99 (6) (2008) 18961903. [7] S. Schiewer, M. Iqbal, The role of pectin in Cd binding by orange peel biosorbents: a comparison of peels, depectinated peels and pectic acid, J. Hazard. Mater. 177 (2010) 899907. [8] I. Aldor, E. Fourest, B. Volesky, Desorption of cadmium from algal biosorbent, Can. J. Chem. Eng. 73 (4) (1995) 516522. [9] F. Bux, F.M. Swalaha, H.C. Kasan, Assessment of acids as desorbents of metalions bound to sludge surfaces, Water SA 21 (4) (1995) 319324. [10] V.P.J. Vilar, C.M.S. Botelho, Copper desorption from Gelidium algal biomass, Water Res. 41 (7) (2007) 15691579.

1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 1 2 3 4 5

Cadmium Concentration C f (mM)

Fig. 6. Comparison of cadmium biosorption isotherms at pH 5.0 of protonated citrus peels and citrus peels desorbed with 0.1 M HNO3 and 0.1 M Ca(NO3 )2 .

248

E. Njikam, S. Schiewer / Journal of Hazardous Materials 213214 (2012) 242248 [18] J.U.K. Oubagaranadin, Z.V.P. Murthy, V.P. Mallapur, Removal of Cu(II) and Zn(II) from industrial wastewater by acid-activated montmorilloniteillite type of clay, C.R. Chimie 13 (2010) 13591363. [19] A.E. Ofomaja, E.I. Unuabonah, N.A. Oladoja, Competitive modeling for the biosorptive removal of copper and lead ions from aqueous solution by Mansonia wood sawdust, Bioresour. Technol. 101 (2010) 38443852. [20] B. Kiran, A. Kaushik, Chromium binding capacity of Lyngbya putealis exopolysaccharides, Biochem. Eng. J. 38 (2008) 4754. [21] A. Hammaini, F. Gonzales, A. Ballester, M.L. Blasquez, J.A. Munoz, Biosorption of heavy metals by activated sludge and their desorption characteristics, J. Environ. Manage. 84 (4) (2007) 419426. [22] R.B. Garcia-Reyes, J.R. Rangel-Mendez, M.C. Alfaro-De la Torre, Chromium (III) uptake by agro-waste biosorbents: chemical characterization, sorptiondesorption studies, and mechanism, J. Hazard. Mater. 170 (2-3) (2009) 845854. [23] S. Schiewer, B. Volesky, Ionic strength and electrostatic effects in biosorption of divalent metal ions and protons, Environ. Sci. Technol. 31 (1997) 24782485. [24] F. Pessoa deFranca, A.P.M. Tavares, A.C. Augusto daCosta, Calcium interference with continuous biosorption of zinc by Sargassum sp. (Phaeophyceae) in tubular laboratory reactors, Bioresour. Technol. 99 (2002) 18961903.

[11] F. Bux, D. Naidoo, H.C. Kasan, Laboratory-scale biosorption and desorption of metal ions using waste sludges and selected acids, S. Afr. J. Sci. 92 (1112) (1996) 527529. [12] K.H. Chu, M.A. Hashim, S.M. Phang, V.B. Samuel, Biosorption of cadmium by algal biomass, adsorption and desorption characteristics, Water Sci. Technol. 35 (7) (1997) 115122. [13] M. Martinez, N. Miralles, S. Hidalgo, N. Fiol, I. Villaescusa, J. Poch, Removal of lead(II) and cadmium (II) from aqueous solution using grape stalk waste, J. Hazard. Mater. 133 (13) (2006) 203211. [14] A. Singh, D. Kumar, J.P. Gaur, Removal of Cu(II) and Pb(II) by Pithopora oedogonia: sorption, desorption and repeated use of the biomass, J. Hazard. Mater. 152 (3) (2008) 10111019. [15] S. Schiewer, A. Balaria, Biosorption of Pb2+ by original and protonated citrus peels: equilibrium, kinetics and mechanism, Chem. Eng. J. 146 (2009) 211219. [16] M. Sciban, B. Radetic, Z. Kevresan, M. Klasnja, Adsorption of heavy metals from electroplating wastewater by wood sawdust, Bioresour. Technol. 98 (2007) 402409. [17] L.D. Fiorentin, D.E.G. Trigueros, A.N. Modenes, F.R. Espinoza-Quinones, N.C. Pereira, S.T.D. Barros, O.A.A. Santos, Biosorption of reactive blue 5G dye onto drying orange bagasse in batch system: kinetic and equilibrium modeling, Chem. Eng. J. 163 (2010) 6877.